COMMENT ⊗   VALID 00004 PAGES
C REC  PAGE   DESCRIPTION
C00001 00001
C00002 00002	\input memo.tex[let,jmc]
C00012 00003	\smallskip\centerline{Copyright \copyright\ \number\year\ by John McCarthy}
C00013 00004
C00023 ENDMK
C⊗;
\input memo.tex[let,jmc]
\title{HOME TERMINALS REVISITED}
%hoter[w87,jmc]		1987 notes for vtss talk on Home computer terminals

	The 1970 paper (computerized in 1976 and last revised in 1980)
advocates developments and makes predictions some of
which have been realized and some not.  It is worthwhile to discuss what
happened and what didn't in order to try to improve prediction and to
advocate what is likely to work.

	The idea that computers could be widely used in the home turned
out to be correct as was the idea that the main uses would concern
communication and information rather than computation per se.  The
biggest error was the failure to anticipate that this use would take
the form of purchase of home computers rather than subscription to
information services delivered through terminals.

	Why do we have home computers rather than merely terminals?
While it is true that microprocessor chips are very cheap in terms of
the cost of pure computation, they are only a small part of a home
computer system, and their computation capability is very slightly
used.  Moreover, home computer systems are much more expensive than
terminals when equipped with disks and printers and enough memory to
run the ambitious programs now available.  Also maintaining a home
computer requires a fair amount of work in making backup tapes, in
filing diskettes and acquiring new software, none of which is required
when using a terminal on a time-sharing computer where this work is
done centrally.  Most people like me whose habits of computer use were
formed with central computers continue to prefer them and haven't
switched.  My wife and I each have a terminal and we've acquired
a laser printer, and all of these are connected to the SAIL computer
at Stanford by a leased telephone line.  The only pressure for change
is that there is some interesting software not available on our main
computer that is available on microcomputers.  I'm also quite certain
that the cost of our use of the main computer is cheaper than amortizing
a microcomputer capable of similar services.

	The reason why microcomputers have won out are commercial,
and they apply to many services besides computers.  When a person
decides to try home computing, he must buy a computer --- at least
almost everyone does rather than rent it.  He, rather than the
manufacturer, takes the risk that he won't use the services much.
Once he has bought the computer, he is more strongly motivated
to learn how to use it than he would be if he were subscribing
to a service where charges are proportional to use.  If the computer
seems unsatisfactory in some way he can always hope that spending
money on an improved system will fix matters.  Advertisers will work
hard to convince him of this.

	There is also the psychological factor of {\it technology
appreciation} previously noticeable in the purchase of cameras, stereo
equipment, cars, boats, airplanes and airplane equipment, etc.  Many
people are observed to be more interested in the equipment than in its
use.  It is often considered to be a psychological abberation, but it
seems to me that owning and becoming familiar with a piece of technology
is psychologically similar to owning and becoming familiar with a piece of
art.  There is the same phenomenon of admiration of the skill of the
makers and tendency to make the features a subject of conversation.

	In general, buying products wins out over renting the use
of systems for reasons associated with the psychology of buyers and
the profits of sellers.  The home computer services proposed in 1970
that require central systems have developed much more slowly than
those that can be packaged as products.

	Consider first the national computerized library proposed
in that paper.  Since about 1970 the magnetic disk technology and
the optical character recognition technology have made it feasible
to make the Library of Congress available in every home and office.
As described in the paper, the creation of such a system will
make freedom of speech far more effective than it is today.  There
is still no political impetus to do it --- maybe there will someday
be a new Andrew Carnegie to endow a computerized public library.

	As an alternative we see the development of optical disks
which can make packages of hundreds of books available to the
home computer user.  However, since each disk will contain a fixed
set of books, the industry hasn't figured out how to get the buyer
to pay for hundreds of books, only a few of which are likely to
be of direct interest to him.

	Centralized computer services have continued to develop
through the era of the home computer.  Indeed home computers that
can also be used as terminals have helped sell the centralized
services.  Even though these services require only a small part
of the power of the home computer, many home computers (and office
personal computers as well) are used mainly as terminals for these
services.

	Another development that hasn't occurred is direct
computer-to-computer mail using the telephone system.  This only
requires the use of common protocols to produce world-wide
electronic mail, but it doesn't provide as good a way of selling
electronic mail as do the proprietary networks.

	The use of home computers for shopping is developing slowly.

	It seems to me that the future will see the developments proposed
and predicted in the 1970 paper.  However, the greater commercial
feasibility of products over services will continue as will the greater
commercial feasibility of competitive private services over utility style
services.  When we'll have the national electronic library is a
random variable depending on when a suitably talented politician
makes it his project.
\smallskip\centerline{Copyright \copyright\ \number\year\ by John McCarthy}
\smallskip\noindent{This draft of HOTER.TEX[W87,JMC]\ TEXed on \jmcdate\ at \theTime}
\vfill\eject\end